Yang Gu

Customize setTimeout

8

If the setTimeout nesting level is greater than 5 and timeout is less than 4ms, timeout will be clamped to 4ms, which hurts the perf. A custom setTimeout is provided...

Compare pose-detection prediction data with another GPU backend

4

We couldn't get CPU prediction data for pose-detection. So to compare the result, we use another GPU backend for reference. To see the logs from the Cloud Build CI, please...

Remove fromPixelsAsync

3

WebGPU no longer relies on async createImageBitmap for fromPixels, so the async version is no longer needed. To see the logs from the Cloud Build CI, please join either our...

fromPixels has different results on CPU and GPU backends

1

Using image, imageBitamp or video as input, fromPixels op follows below steps: 1) A 2D canvas is created To get the drawing context of the canvas, willReadFrequently is set to...

[js/webgpu] Refactor trace

1

Current trace impl relies on unmangled JS to get the function names, which causes some inconvenience and perf impact. This refactoring removes this dependence.

Add missing options in pipeline API

3

This is to add missing options in pipeline API to align with from_pretrained API. Bug #883 can be fixed with this. That model has a separate data file, so use_external_data_format...

### Description ### Motivation and Context

Add vscode settings

3

Maybe this is more friendly for vscode users like me.

[BUG] Could not run DirectML models on Intel laptop

6

**Describe the bug** My laptop doesn't have any discrete GPU, but an Intel integrated GPU (Meteor Lake). However, I have 32G system memory, and Intel iGPU may use up to...